Audio overview

A spoken overview of this page, for anyone who prefers to listen.

What moves the race — votes versus money

See which forces weigh most on the win call and which weigh most on the money. Open any state on the map for the local picture.

Group charts

Pooled all-50 · New England · Trump 2024 red · Harris 2024 blue. PNGs embedded below.

Relative beta importance win versus Wealthy all fifty states
All 50 states (pooled) · t=2026-10-24
Relative beta importance New England
New England · CT ME MA NH RI VT · t=2026-10-24
Relative beta importance Trump 2024 red states
Red states (Trump 2024) · t=2026-10-24
Relative beta importance Harris 2024 blue states
Blue states (Harris 2024) · t=2026-10-24

Fifty states — pick a state

Click a state on the map (blue = available, red = selected). Arrow keys move between states when the map has focus. Each panel shows that state’s explainability card — state-win weights, plain language, and the Wealthy model — with the receipt table underneath.

United States — select a state Alaska Alabama Arkansas Arizona California Colorado Connecticut Delaware Florida Georgia Hawaii Iowa Idaho Illinois Indiana Kansas Kentucky Louisiana Massachusetts Maryland Maine Michigan Minnesota Missouri Mississippi Montana North Carolina North Dakota Nebraska New Hampshire New Jersey New Mexico Nevada New York Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Virginia Vermont Washington Wisconsin West Virginia Wyoming

Selected: —

AK — Alaska

State explainability card for Alaska: state-win LOO weights, plain language, and Wealthy model, October 24, 2026
State explainability card · state-win LOO coefficient stability, plain language, and Wealthy model · AK · t=2026-10-24

Charts Explainability card above · campaign-finance receipt figures for this state are listed below

Vote Statewide call: GOP (p_Dem=3%) · 2026 money shape: balanced

2026 race ratings (AI-assisted update, 2026-10-30 21:06 EDT): Senate (Sullivan): Cook Toss Up · DDHQ 60% D to win; Governor (open seat): Cook Lean D. Sources: Cook Senate (Oct 30), Cook Governor (Oct 29), DDHQ Senate (Oct 29). The statewide call above comes from the Oct 24 model snapshot, not from these ratings.

Money Receipts by cycle (campaign-finance weball)

Showing top 3 candidates by weball receipts (of 3 in DB).

CycleCandidateTotal receiptsIndividualPAC (other pol. cmte)
2024BEGICH, NICHOLAS III$2.81M$2.31M$319k
2026BEGICH, NICHOLAS III$6.44M$3.31M$1.24M
2024MURKOWSKI, LISA$689k$356k$277k
2026MURKOWSKI, LISA$1.38M$939k$377k
2024SULLIVAN, DAN$1.89M$915k$524k
2026SULLIVAN, DAN$11.51M$5.71M$2.28M

Money flow PAC & individual money

PAC money was about $1.12M of $5.39M total receipts in 2024 and $3.90M of $19.33M in 2026. Direct committee→candidate 24K (non-memo) ran about $1.21M (2024) / $3.83M (2026). Itemized individual share from Alaska: 66.4% (2024) / 51.5% (2026).

SULLIVAN, DAN: Cycle-to-date weball receipts ~$11.51M, with PAC direct 24K $525k → $2.30M vs 2024. the public campaign-finance data feed by-size individual totals are not loaded in this 50-state snapshot.

Ballot Spotlight candidates

SULLIVAN, DAN (Senate AK)

  • In 2024, $915k of $1.89M came from individuals; PAC (other pol. cmte) $524k.
  • In 2026, $5.71M of $11.51M came from individuals; PAC (other pol. cmte) $2.28M.

BEGICH, NICHOLAS III (Senate AK)

  • In 2024, $2.31M of $2.81M came from individuals; PAC (other pol. cmte) $319k.
  • In 2026, $3.31M of $6.44M came from individuals; PAC (other pol. cmte) $1.24M.

MURKOWSKI, LISA (Senate AK)

  • In 2024, $356k of $689k came from individuals; PAC (other pol. cmte) $277k.
  • In 2026, $939k of $1.38M came from individuals; PAC (other pol. cmte) $377k.

Teleprompt Short talking points — money & predicted voting

TELEPROMPT — Alaska (AK) money & midterm call

Open with the model set’s statewide call: GOP favored at about 3% Dem win probability on public campaign-finance + roster features — not a poll. Then the money shape: Alaska federal candidates look balanced in 2026 weball (about $3.90M PAC / $9.96M individual of $19.33M total). Spotlight SULLIVAN, DAN for one concrete receipt picture. Close civic: these are public contribution patterns and a transparent forecast model call — not evidence of wrongdoing, and not a substitute for Cook or election-night counts.

Source: states/AK/the finance database fec_candidate_summary + fec_committee_to_candidate 24K. Patterns ≠ wrongdoing.

Fifty-state briefing cards

Click a state on the map (blue is available, red is selected). Use the arrow keys when the map is focused. Each state shows its explainability card, the money picture, and short talking points. Snapshot date: October 24, 2026 (model cards). Race ratings refreshed Oct 30, 2026, 9:06 p.m. EDT (AI-assisted from Cook Political Report and Decision Desk HQ).

United States — select a state Alaska Alabama Arkansas Arizona California Colorado Connecticut Delaware Florida Georgia Hawaii Iowa Idaho Illinois Indiana Kansas Kentucky Louisiana Massachusetts Maryland Maine Michigan Minnesota Missouri Mississippi Montana North Carolina North Dakota Nebraska New Hampshire New Jersey New Mexico Nevada New York Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Virginia Vermont Washington Wisconsin West Virginia Wyoming

Selected: —

AK — Alaska

State explainability card for Alaska: state-win LOO weights, plain language, and Wealthy model, October 24, 2026
Charts State explainability card · state-win LOO coefficient stability, plain language, and Wealthy model · AK · t=2026-10-24

Vote Statewide call: GOP (p_Dem=3%) · 2026 money shape: balanced

2026 race ratings (AI-assisted update, 2026-10-30 21:06 EDT): Senate (Sullivan): Cook Toss Up · DDHQ 60% D to win; Governor (open seat): Cook Lean D. Sources: Cook Senate (Oct 30), Cook Governor (Oct 29), DDHQ Senate (Oct 29). The statewide call above comes from the Oct 24 model snapshot, not from these ratings.

Money Receipts by cycle (campaign-finance weball)

Showing top 3 candidates by weball receipts (of 3 in DB).

CycleCandidateTotal receiptsIndividualPAC (other pol. cmte)
2024BEGICH, NICHOLAS III$2.81M$2.31M$319k
2026BEGICH, NICHOLAS III$6.44M$3.31M$1.24M
2024MURKOWSKI, LISA$689k$356k$277k
2026MURKOWSKI, LISA$1.38M$939k$377k
2024SULLIVAN, DAN$1.89M$915k$524k
2026SULLIVAN, DAN$11.51M$5.71M$2.28M

Money flow PAC & individual money

PAC money was about $1.12M of $5.39M total receipts in 2024 and $3.90M of $19.33M in 2026. Direct committee→candidate 24K (non-memo) ran about $1.21M (2024) / $3.83M (2026). Itemized individual share from Alaska: 66.4% (2024) / 51.5% (2026).

SULLIVAN, DAN: Cycle-to-date weball receipts ~$11.51M, with PAC direct 24K $525k → $2.30M vs 2024. the public campaign-finance data feed by-size individual totals are not loaded in this 50-state snapshot.

Ballot Spotlight candidates

SULLIVAN, DAN (Senate AK)

  • In 2024, $915k of $1.89M came from individuals; PAC (other pol. cmte) $524k.
  • In 2026, $5.71M of $11.51M came from individuals; PAC (other pol. cmte) $2.28M.

BEGICH, NICHOLAS III (Senate AK)

  • In 2024, $2.31M of $2.81M came from individuals; PAC (other pol. cmte) $319k.
  • In 2026, $3.31M of $6.44M came from individuals; PAC (other pol. cmte) $1.24M.

MURKOWSKI, LISA (Senate AK)

  • In 2024, $356k of $689k came from individuals; PAC (other pol. cmte) $277k.
  • In 2026, $939k of $1.38M came from individuals; PAC (other pol. cmte) $377k.

Teleprompt Short talking points — money & predicted voting

TELEPROMPT — Alaska (AK) money & midterm call

Open with the model set’s statewide call: GOP favored at about 3% Dem win probability on public campaign-finance + roster features — not a poll. Then the money shape: Alaska federal candidates look balanced in 2026 weball (about $3.90M PAC / $9.96M individual of $19.33M total). Spotlight SULLIVAN, DAN for one concrete receipt picture. Close civic: these are public contribution patterns and a transparent forecast model call — not evidence of wrongdoing, and not a substitute for Cook or election-night counts.

The Fight for a Vibrant Democracy

All of us know the rocky and tortuous road that American democracy has traveled since the founding of our nation, when colonial revolutionaries took on and defeated the king of England and the enormous power of the British Empire. Determined that the decisions that impacted their lives should be made by the people themselves, and not by an autocratic monarch three thousand miles away across the Atlantic Ocean, the revolutionaries fought to create a new kind of society, a democratic society, unique in the world of its time.

“Defeating Oligarchy,” Our Revolution, 2016, by Bernie Sanders.

Kit GLM model (Oct 30, 2026 rescore)

Model estimate and opinion, not fact. These numbers come from Sam Castillo’s own Kit GLM. They sit beside the Cook Political Report ratings and the AI projection on this page and do not replace either one.

Sam Castillo in a charcoal suit and blue tie, arms crossed
Sam CastilloFounder & AI Engineer · Predictive Insights AIActuary and UMass Amherst mathematician who built and runs the Kit GLM.
21 / 29states lean Democratic / Republican
≈199average Democratic House seats
≈18%chance of a Democratic House majority (218)
0.03 ptslargest move in any state since Oct 24

50-state win chances

The model gives Democrats the better chance in 21 states and Republicans the better chance in 29.

Bar chart of the 21 states where the Kit GLM gives Democrats the better chance, from New Hampshire near 95% down to Colorado about 56% and Michigan about 51%, with Oct 24 and Oct 30 nearly identical
Leaning Democratic: 21 states.
Bar chart of the 29 states where the Kit GLM gives Republicans the better chance, led by Minnesota at about a 35% Democratic chance, with a callout noting the model leans Republican in Minnesota
Leaning Republican: 29 states. Minnesota is the closest.

House seats

A 20,000-run simulation with a shared national swing gives an average of about 199 Democratic seats. The middle 50% of runs fall between 185 and 213 seats and the 95% range is 157 to 240. Democrats reach the 218 needed for a majority in about 18% of runs.

Histogram of simulated Democratic House seats centered near 199, with a dark band for the middle 50% from 185 to 213, a light band for the 95% range from 157 to 240, a dashed line at the 218 majority, and a dotted line at 229 marking the separate AI projection
Simulated Democratic House seats. The 229 line is the separate AI projection, not this model.

“We should inquire into the Tempers of Men, in order to form a Judgment in what Manner the publick Trusts to be reposed in them will be executed.”

Senate seats

The Kit GLM gives Democrats (counting the two independents who caucus with them) an average of 45.6 Senate seats and about a 0.5% chance of reaching 51.

Histogram of simulated Democratic caucus Senate seats, including two independents, centered near 46, with a dark band for the middle 50% from 44 to 47, a light band for the 95% range from 42 to 49, a dotted line at a 50-seat tie and a dashed line at the 51-seat majority; about a 0.5% chance of 51 or more
Simulated Democratic caucus Senate seats. A 50–50 tie goes to the vice president.

Chamber odds at a glance

Kit GLM model estimate, Oct 30 rescore

House: 18.4% chance of a Democratic majority (218+ seats); average about 199 Democratic seats.
Senate: 0.5% chance of 51+ Democratic caucus seats; average 45.6 seats.

These are the Kit GLM’s own odds. The House and Senate donuts further down this page show the separate Oct 24 AI projection and Cook ratings, not this model.

What drives the state win model nationally

Pooled logistic coefficients across all 50 states, largest first:

  1. House Democratic share: 2.82
  2. Money lean (Democratic minus Republican fundraising): 1.71
  3. A Democrat holds the Class 2 Senate seat: 0.42
  4. ActBlue Democratic share of small donations: −0.14

These are raw coefficients and each input is on a different scale, so treat the order as a guide, not an exact measure.

Dot chart of state win model coefficients: House Democratic share about 2.8, money lean about 1.7, Democrat-held Class 2 Senate seat about 0.4, and ActBlue share slightly below zero, with ranges across 50 leave-one-out refits and the Massachusetts fit marked
What moves the state win model.
Massachusetts model card: bar chart of each input’s push on the score, led by money lean and House seat share, and a table showing a 95.4% Democratic win chance on both Oct 24 and Oct 30
Massachusetts: 95.4% on both Oct 24 and Oct 30.

“Let us contemplate our forefathers and posterity; and resolve to maintain the rights bequeath’d to us from the former, for the sake of the latter.”

Quotations are Samuel Adams’s own words as printed in The Writings of Samuel Adams, ed. Harry Alonzo Cushing: the “Candidus” essay, Boston Gazette, Oct 14, 1771 (vol. 2), and his letter to James Warren, Oct 24, 1780 (vol. 4).

What changed on Oct 30

The Oct 30 rescore re-ran the Oct 24 models on refreshed data without refitting them. Almost nothing changed: the largest move in any state was 0.03 points, because the FEC filings behind the money inputs mostly run through Oct 14.

Method, in plain terms

Each state gets a regularized logistic regression for the win call, plus Poisson and Gamma models for seats and money. The training labels are House-majority stand-ins, not real past state results. The model uses no polls and no early-vote data. Treat every number here as a model estimate and opinion, not fact.

Our forecast beside the October 24, 2026 AI projection

This reading sits beside the Cook Political Report and the October 24 AI chamber snapshot. It does not replace either. Same day — October 24, 2026 — different questions.

Source What it is
Cook Political Report reading Rating-style Senate map; current Dem caucus 47; Cook-style majority about 51–52 D / 48–49 R from about +2 to +5. Left intact on the DC page.
AI projection (AI snapshot 18:35 ET) National chamber forecast: House 229 D / 206 R (net Dem +13); Senate Dem caucus 49–51 with P(Dem Senate)=0.62; Senate Dem intervals 50% [49, 53], 75% [48, 54], 90% [46, 56].
Model set forecast model 50-state Follow the Money models: yes-or-no (win/lose) state-win the forecast∈yes or no; count House seat split inside each state; no fitted national Senate chamber model. The Oct 30 rescore adds a Kit GLM Senate overlay: a simulation built from the state win models (not a fitted chamber model) that gives about a 0.5% chance of 51+ Democratic caucus seats.

Model set snapshot: follow-the-money_50states/out/america_criteria_glm_predictions.csv + out/state_win_predictions.csv.
AI snapshot: midterms2026/data/forecast_snapshot.json (House 229/206; Senate 51/49 documented with the 47+2=49 caveat).

House — Kit GLM versus the AI projection

National rollup does not assume independence across states. State μ_s stay; national (D(\theta)=\sum_s n_s\,\sigma(\mathrm{win-probability link}(p_s)+\theta)), (\theta\sim N(0,\tau^2)). Independent (\mathrm{count}(\sum\mu)) is kept only as a rejected benchmark (scripts/house_seats_national.py).

Kit GLM (shared-shock national) AI projection Gap
Dem seats (point) 199.4 (round 199) 229 −30
GOP seats (point) 232.6 (round 233) 206 +27
Chamber size in that split 432 (50-state D+R only) 435 −3
Control (point) GOP (199 < 218) Dem (229 ≥ 218) disagree
P(Dem House) ≈ 0.18 (shared shock MC) 0.82 still large
sd(Dem) ≈ 21.1 (shared); indep Poisson sd≈14.1 not used — wider under dependence
50% interval (Dem) ≈ [185, 213] [225, 233] no overlap
75% ≈ [175, 223] [220, 237] touch
90% ≈ [164, 234] [216, 241] partial overlap

The quoted bands [49, 53] / [48, 54] / [46, 56] are Senate Democratic seats, not House.

Why they differ: the forecast model reallocates each state’s current House seats by that state’s win score — not challenger pickups / national generic ballot. Sitting 50-state D+R in the model set is 214 D / 218 R (432).

Senate

The model set does not fit a Senate-chamber forecast model. Overlay below uses Class-2 incumbency from the members DB plus the state-win the forecast (trained on House-majority proxy — not a Senate-race model).

Current 119th (model set roster): Dem 45 + Independent 2 = caucus 47; GOP 53.
Class 2 (up in 2026): 13 D / 20 R. Held (Class 1 D+I + Class 3 D) = 34.

Model set overlay (Class-2 + state-win) AI projection Gap
Dem caucus point hard the forecast 46 (MN Class-2 flip D→R; no R→D); E[p] 45.6 49–51 (raw model printed 51/49; 47+2=49 if net +2) about −3 to −5
GOP the forecast 54 49–51 opposite lean
P(Dem Senate) ≈ 0.006 (P(34+S≥50), S = sum of 33 Class-2 Bernoullis, E[S]=11.55) 0.62 disagree
50% Dem seats model set E ≈ 46, sd ≈ 1.6 → roughly [44, 47] [49, 53] no overlap
75% ~[43, 48] [48, 54] touch at 48
90% ~[43, 48] [46, 56] model set sits at the low end of the AI 90% band

Only Class-2 hard flip vs current: Minnesota (the forecast GOP, p_dem=0.35). Competitive Class-2 keeps: CO (0.56), MI (0.51). No Republican Class-2 state is called Dem.

Cook (beside, not replaced): start at 47; majority path +2 to +5 → about 51–52 D. The AI 0.62 and 49–51 range sit on that side of 50. The model set overlay sits below 50 and is not a Cook substitute.

A state win call is not a full chamber forecast

Model set state-win (50 states) Meaning
DEM 21 / GOP 29 Statewide midterm call, not House or Senate seats
Agree House-majority proxy 45/50

Do not add 21+29 to 229/206 or to 49–51.

How to read the three together

  1. Cook — qualitative ratings and a 47-seat caucus baseline.
  2. AI projection — national House/Senate chamber snapshot for October 24 (House internally consistent; Senate point 51 vs +2-from-47 needs the 49–51 caveat).
  3. Model set forecast model — money + roster state models. House count is incumbent-seat allocation (GOP-leaning vs AI). Senate chamber probability is an overlay, not a fitted Senate forecast model.
cd /home/sandcastle/midterms2026/follow-the-money_50states
python3 scripts/state_win_classifier.py --date 2026-10-24 --fit
python3 scripts/glm_families.py --date 2026-10-24

Machine copy of this comparison: model set out/glm_vs_oct24_projection.json.

Weight explainability (forecast model open model): docs/glm-coefficient-explainability.md · charts in media/glm-explain/ (model set out/explain/).

Open weights versus a closed AI snapshot

October 24, 2026. The Follow the Money fifty-state models publish every weight, so a campaign can see what is driving the call. The October 24 AI chamber projection does not publish those weights. Cook ratings sit beside both; neither replaces Cook.

Charts

File What it shows
01_state_win_coefficients Pooled yes-or-no (win/lose) win-probability link weight — sign (+ Dem / − Dem) and magnitude
02_wealthy_gamma_coefficients Wealthy positive-dollar log-link weight on campaign-finance vitality $
03_state_win_loo_coef_spread stability check on each weight across 50 states (stability of sign)
04_blackbox_vs_glm_explainer Side-by-side: closed AI snapshot vs model set open model
05_relative_importance_win_vs_wealthy Which factors matter most ranks, win vs Wealthy (all-50 pooled; alias of split 00_all50_combined)

Which factors matter most splits (out/explain/relative_importance/)

Chart Scope
00_all50_combined Pooled / all 50
by_state/<ST>.png stability check on each weights for each state (50 charts)
new_england Mean |weight| over CT, ME, MA, NH, RI, VT
red_trump_2024 Mean |weight| over states Trump won in 2024
blue_harris_2024 Mean |weight| over states Harris won in 2024

Calculations: coefs_win_vs_wealthy.csv · coefs_groups.json.
Rebuild splits: python3 scripts/plot_relative_importance_splits.py --date 2026-10-24

Briefing cards (all 50 states)
PNGs out/explain/cards_by_state/<ST>.png
HTML tabset glm-explainer.html#state-cards (same page) · cards_by_state/index.html
Rebuild python3 scripts/plot_state_explainability_cards.py --date 2026-10-24 --html

Each card: state-win stability check on each weights bars · plain-language explainability panel · Wealthy weight row.

Paths:

How to read a weight

Meaning
Sign + raises Dem win odds or expected $; − lowers them
Magnitude Larger |weight| = stronger partial effect on the link scale (log-odds or log-$)
Money lean Follow-the-money signal (Dem vs GOP receipts) — inspectable, not buried

Largest state-win drivers (pooled): House Dem share (+2.82) · Money lean (+1.71) · Class-2 Dem (+0.42). ActBlue share is small and auxiliary (−0.14).

Rebuild

cd /home/sandcastle/midterms2026/follow-the-money_50states
python3 scripts/plot_glm_explain.py --date 2026-10-24
# knit MD → HTML with Data Tea Party hero + politician_icons
cd /home/sandcastle/midterms2026
python3 scripts/knit_md_html.py
# → site/glm-explainer.html  ·  docs/glm-explainer.html  ·  product/glm-explainer.html
State-win logit β — sign and magnitude
State-win win-probability link weight — sign and magnitude
Perplexity black box vs kit GLM explainer
closed AI snapshot vs model set forecast model explainer
Wealthy Gamma coefficients
Wealthy positive-dollar weights
Relative |β| win vs Wealthy
Which factors matter most win vs Wealthy

What kind of model fits which question

Pick the tool that matches the question: a yes-or-no win, a seat count, campaign dollars, or a rate between zero and one. That choice is what keeps the forecast honest and the weights open for a briefing.

Notation

Symbol Meaning
t prediction date (daily snapshot)
today’s inputs date matrix + inputs at t
y outcome
the forecast = f(today’s inputs) predicted mean
the leftover error residual when using additive identity-normal-curve form

Additive y = f(X) + the leftover error with the leftover error ~ normal-curve is the right default only when y is roughly unbounded and continuous. For other supports, use a forecast model: the model links the average outcome to the weighted inputs with the likelihood below.

Family selection (best default by support of y)

Family y support Assumption Use here
Yes-or-no win Win or lose Probability of a Democratic win State win call
score model any real number score equals weighted inputs plus leftover error Greatness score, margins
Seat-count model 0, 1, 2, … seats Expected seats from a count model House seats inside a state
Seat-count model 0, 1, 2, … seats Expected seats from a count model House seats inside a state
money model positive amounts positive continuous, spread grows with the square of the average Wealthy $ proxies (receipts)
Score model Any score or margin Straight-line score with leftover error Institutional score, margins
Share-or-rate model Between 0% and 100% Rates and shares only Participation or health-style rates

Do not treat binary wins as y = Xβ + ε_Gaussian as the primary model (straight-line probability shortcut only as a diagnostic).

How the numbers are built — wins, seats, and money

Every number on this page answers a different kind of question. A win call is not a seat count. A seat count is not a fundraising total. Below is the plain structure behind those choices — the same structure that produces the open weights on the state cards.

Core forecast model all models

How inputs become a forecasted average.

\[ g(\mu) = \mathbf{x}^{\top}\boldsymbol{\beta}, \qquad \mu = \mathbb{E}[y \mid \mathbf{x}] = g^{-1}(\mathbf{x}^{\top}\boldsymbol{\beta}) \]

Each day’s inputs — leans, timing, and related measures — are weighted and combined. Those open weights are what the state cards and “what matters most” charts show.

yes-or-no win model state win

For a single race: Democrat wins or does not.

\[ y \in \{0,1\}, \quad p = \Pr(y=1\mid\mathbf{x}) \]
\[ \operatorname{logit}(p) = \log\frac{p}{1-p} = \mathbf{x}^{\top}\boldsymbol{\beta} \]
\[ p = \sigma(\mathbf{x}^{\top}\boldsymbol{\beta}) = \frac{1}{1+e^{-\mathbf{x}^{\top}\boldsymbol{\beta}}} \]

This is the standard win-probability model. Do not treat a binary race as a plain straight-line score as the main call.

score model scores and margins

For scores and margins that can move up or down without a hard floor or ceiling.

\[ y \mid \mathbf{x} \sim \mathcal{N}(\mu,\sigma^{2}), \quad g(\mu)=\mu \]
\[ f(y\mid\mu,\sigma^{2}) = \frac{1}{\sqrt{2\pi\sigma^{2}}} \exp\!\left(-\frac{(y-\mu)^{2}}{2\sigma^{2}}\right) \]
\[ y = \mathbf{x}^{\top}\boldsymbol{\beta} + \varepsilon, \quad \varepsilon \sim \mathcal{N}(0,\sigma^{2}) \]

straight-line fit is best fit here. Default only when y is roughly continuous and unbounded.

seat-count model House seats

For expected House seats inside a state — whole numbers only.

\[ y \in \{0,1,2,\ldots\}, \quad y \mid \mathbf{x} \sim \mathrm{Poisson}(\mu) \]
\[ \Pr(Y=k) = \frac{e^{-\mu}\mu^{k}}{k!}, \quad \log\mu = \mathbf{x}^{\top}\boldsymbol{\beta} \]
\[ \mathbb{E}[Y] = \mathrm{Var}(Y) = \mu = e^{\mathbf{x}^{\top}\boldsymbol{\beta}} \]

In the basic count model, the average and the spread match. If seats jump around more than that, use the wider-spread count model. A national House total should not treat every state as independent.

more spread than a simple count model count wide-spread counts

For seat or event counts when results swing more than a simple count model allows.

\[ \mathbb{E}[Y] = \mu, \qquad \mathrm{Var}(Y) = \mu + \frac{\mu^{2}}{\theta} \]
\[ \log\mu = \mathbf{x}^{\top}\boldsymbol{\beta} \]

When the extra-spread control gets very large, this model behaves like the basic count model. Use it when seat or event counts fan out more than a simple average would suggest.

money model campaign money

For campaign dollars and other positive money measures.

\[ y > 0, \quad \log\mu = \mathbf{x}^{\top}\boldsymbol{\beta}, \quad \mathrm{Var}(Y) \propto \mu^{2} \]
\[ f(y\mid\mu,\nu) = \frac{1}{\Gamma(\nu)} \left(\frac{\nu}{\mu}\right)^{\nu} y^{\nu-1} e^{-\nu y/\mu} \]

Built for right-skewed dollars — most states moderate, a few very large. Standard choice for campaign-receipt style measures.

Inverse normal-curve very skewed money

For money that is even more top-heavy than a standard dollar model.

\[ y > 0, \quad \log\mu = \mathbf{x}^{\top}\boldsymbol{\beta} \]
\[ f(y\mid\mu,\lambda) = \sqrt{\frac{\lambda}{2\pi y^{3}}} \exp\!\left( -\frac{\lambda(y-\mu)^{2}}{2\mu^{2} y} \right) \]

Use when a few states dominate the money picture. Compare stability against the standard dollar model on the same inputs.

share-or-rate model rates and shares

For shares and rates that must stay between zero and one.

\[ y \in (0,1), \quad \operatorname{logit}(\mu) = \mathbf{x}^{\top}\boldsymbol{\beta} \]
\[ f(y\mid\mu,\phi) = \frac{\Gamma(\phi)}{\Gamma(\mu\phi)\Gamma((1-\mu)\phi)} y^{\mu\phi-1}(1-y)^{(1-\mu)\phi-1} \]

For participation or health-style rates that must stay between 0% and 100%. Handle exact zeros or ones with care.

Why not one normal-curve for everything?

Match the tool to the question.

A one-size-fits-all straight line can invent negative seats, odds outside 0–100%, or negative dollars. Matching the model to the kind of outcome keeps the forecast in the real world — and keeps every weight open for a campaign brief.

Criteria (Sam Castillo)

Massachusetts roots: grandfather a Boston firefighter; three brothers served in the military; worked for the State of Massachusetts. Product criteria are civic and measurable:

  1. US Democratic Election 2026 → fitted positive-dollar / IG stability check on 2024–2026 campaign-finance state receipts (scripts/america_wealthy.py); 2027–2032 vitality forecasts; saved models under out/models/america_wealthy_*.pkl.
  2. Make America Healthy Again → share-or-rate on between zero and one participation/health-style rates (national CDC feeds to join later; MA registration/population already in model set).
  3. Make America Great Again → yes-or-no (win/lose) win call + normal-curve institutional score from today’s inputs.

Run

cd /home/sandcastle/midterms2026/follow-the-money_50states
python3 scripts/state_win_classifier.py --date 2026-10-24   # Bernoulli win
python3 scripts/america_wealthy.py --date 2026-10-24        # Wealthy Gamma/IG × 50
python3 scripts/glm_families.py --date 2026-10-24           # full family suite (pulls Wealthy)

Outs: out/america_wealthy_predictions.csv, out/glm_family_catalog.json, out/america_criteria_glm_predictions.csv, out/models/america_wealthy_bundle.pkl

Chamber comparison vs the October 24 AI / Cook snapshot (does not replace it): docs/glm-vs-oct24-ai-projection.md.

House national total: do not assume independence across states. scripts/house_seats_national.py — shared win-probability link shock; outs out/house_seats_national_dependent.json. Independent (\mathrm{count}(\sum\mu)) is a rejected benchmark only.

God bless America.

Real Tax Reform

For the past fifty years, Wall Street banks, large profitable corporations, and the billionaire class have rigged the tax code to redistribute wealth and income to the richest and most powerful people in this country. At a time of massive wealth and income inequality, when major corporation after major corporation pay nothing in federal income taxes and many CEOs enjoy an effective tax rate that is lower than their secretaries’, we need progressive income tax reform based on the ability to pay.

Real Tax Reform, page 263, Our Revolution, Bernie Sanders.

Independence 250 — Washington, D.C.

House and Senate outlook, how the forecast is built, and a photo journal from one Monday on the National Mall — August 17, 2026.

🇺🇸 ★

Independence 250

250 Years of Liberty • July 4, 1776 – July 4, 2026

Washington, D.C., is the capital of the United States.

★ ★ ★ When in the Course of human events... ★ ★ ★

One Nation, Under God, with Liberty and Justice for All

Read the Monday journal → Predict Election

Bakery box of red, white, and blue donuts with a center donut reading Sam and Make America Wealthy Again Poster-style stack of red, white, and blue donuts reading Sam and Make America Wealthy Again
Make America Wealthy Again

ELECTION PREDICTIONS

Twice-daily reading of public polls, ratings, and prediction markets for the November 3, 2026 United States midterm elections. This page refreshes at 6:00 a.m. and 6:00 p.m. Eastern through Election Day.

This is a model forecast assembled from public sources. See Sources / Notes for citations.

Where Does Your Ballot Actually Go?

Where Does Your Ballot Actually Go? How Votes Get Counted, Step by Step (AI-Narrated)

Updated: 2026-10-30 18:08 EDT

Four days before the November 3, 2026 midterms, the major forecasts and prediction markets favor Democrats to win the House by a wide margin and give them the edge in a closer Senate fight.

House of Representatives

  • Decision Desk HQ (updated Oct 29) gives Democrats an 80% chance of winning the House, with a projected 238 seats.
  • Silver Bulletin (Nate Silver) gives Democrats a 95.1% chance and projects 240 seats, a net gain of about 25 (as reported Oct 30).
  • Cook Political Report (Oct 30) rates about 222 seats as leaning or safe Democratic and 195 as leaning or safe Republican, with the rest (18) toss-ups. Cook says the most likely outcome is Democrats holding 227 to 240 seats.
  • Polymarket prices Democrats at about 93.5% and Kalshi at about 92.5% to win the House (Oct 30).

Senate

  • Decision Desk HQ gives Democrats a 60% chance of winning the Senate, with a projected 52 seats.
  • Silver Bulletin (Nate Silver) puts Democrats at about 78–79%, projecting a net gain of about six seats (Oct 30).
  • Cook Political Report (Oct 30), counting seats not up this year, rates about 47 seats as leaning or safe Democratic and 46 as leaning or safe Republican, with the rest (7) toss-ups: Alaska, Iowa, Kansas, Maine, Michigan, Ohio and Texas. Cook calls Democrats slight favorites.
  • Polymarket prices Democrats at about 72.5% and Kalshi at about 69.5% to win the Senate (Oct 30).

Bottom Line

Democrats are strongly favored to win the House, with forecasts centered on roughly 227–240 Democratic seats, and are favored but not safe in the Senate, with projections around 51–53 Democratic seats. Several Senate toss-ups are within a normal polling error.

Important caveat: These are forecasts and probabilities, not election results. Actual outcomes depend on turnout, candidate quality, late-breaking events, and how competitive races break on Election Day.

Previous update: Oct 28, 2026, 6:13 p.m. EDT

Earlier snapshot, kept unchanged for comparison.

Updated: 2026-10-28 18:13 EDT

The 6:00 p.m. note used Cook’s ratings alone. This reading keeps that House map and adds Decision Desk HQ’s October 28 seat model, last updated at 10:19 a.m. Eastern.

U.S. House of Representatives

On October 23, Cook moved 19 more House races toward Democrats and wrote that Republicans’ path to keeping the House has all but evaporated. Democrats are favored in 220 districts (Solid, Likely, or Lean), Republicans in 198, and 17 races are toss-ups. Democrats currently hold 215 seats. Cook’s likeliest outcome is a flip of 12 to 22 Republican-held seats, about 227 to 237 Democratic seats. Winning none of the 17 toss-ups would still leave Democrats on the 220 seats where they are already favored, past the 218 needed for a majority. Winning all 17 would reach 237.

Decision Desk HQ’s October 28 model gives Democrats an 80 percent chance of winning the House and projects 238 seats. That point is one seat above Cook’s rating-map ceiling of 237.

U.S. Senate

Cook’s Senate ratings page, updated October 20, shows seven toss-ups. Of the 35 seats on the ballot, Cook rates 13 Solid, Likely, or Lean Democratic (9 solid, 2 likely, 2 lean) and 15 Solid, Likely, or Lean Republican (12 solid, 2 likely, 1 lean). Thirty-four Democratic-held seats and 31 Republican-held seats are not up this year.

Decision Desk HQ’s October 28 model gives Democrats a 58 percent chance of winning the Senate and projects 51 seats. The chambers do not move together. DDHQ’s joint chart reads 56 percent for a Democratic House and Senate, 23 percent for a Democratic House with a Republican Senate, 19 percent for Republican control of both, and 2 percent for a Democratic Senate with a Republican House. Those two Democratic-House slices add to 79, one point under the 80 percent House headline, because each slice is rounded on its own.

Bottom line

House control is the settled part of the forecast. Cook’s favored column is already a majority, and DDHQ’s point forecast is a larger one, at an 80 percent chance and 238 seats. Senate control is a lean, not a lock: 58 percent, with a 51-seat point. The result that still changes the story is split control, a Democratic House and a Republican Senate, which DDHQ’s joint chart puts at 23 percent.

Defeating Oligarchy

"We should be forever grateful to Jefferson, Adams, Paine, Washington, Madison, Franklin, Hamilton, and all the other revolutionary leaders who founded our country. With incredible courage, they put their lives on the line to create a very different kind of society. Democracy is the right of a free people to control their destiny. Not kings or queens or czars, but ordinary people who come together in a peaceful manner in order to determine the future of their society. Democracy means that the government belongs to all of us and that it is our inherent right to elect people who will represent our interests."1

"After all, this is what our Declaration of Independence proclaims when it profoundly states: 'We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness. That to secure these rights, Governments are instituted among Men, deriving their just powers from the consent of the governed.' Their just powers from the consent of the governed."2

"Democracy is about one person, one vote. It's about all of us coming together to determine the future of our country. It is not about a handful of billionaires buying elections, or governors suppressing the vote by denying poor people or people of color the right to vote. Our job is to stand together to defeat the drift toward oligarchy and create a vibrant democracy."1

Why it matters on November 3: one person, one vote only works if everyone can cast one and every ballot gets counted. Make your plan to vote.

1 Bernie Sanders, Our Revolution: A Future to Believe In (New York: Thomas Dunne Books, 2016).
2 Sanders, Our Revolution, 186, quoting the Declaration of Independence (1776).

AI Midterm Forecast

October 24, 2026 projection. This sits beside the Cook Political Report reading above; it does not replace it.

AI projection

House control leans Democratic (about 229 Democratic seats / 206 Republican; net Democratic House gain about +13). Senate control also leans Democratic, with the Democratic caucus projected in the 49–51 range (see Senate Forecast).

ProbabilityValue
P(Democrats win House)0.82
P(Democrats win Senate)0.62

Share at a glance — blue for Democrats, red for Republicans. Hover or focus a chart for sources; click to pin the list. Two light desk charts (Diet Coke and Gummy Bears) are not race forecasts. See also how the forecast is built.

Donut chart: 82 percent chance Democrats win the House, 18 percent Republicans. Hover or focus for sources.
House control chance — Democrats 82%, Republicans 18%
Donut chart: 62 percent chance Democrats win the Senate, 38 percent Republicans. Hover or focus for sources.
Senate control chance — Democrats 62%, Republicans 38%
Donut chart: projected House seat share Democrats 53 percent (229 seats), Republicans 47 percent (206 seats). Hover or focus for sources.
House seats — Democrats 229 (53%), Republicans 206 (47%)
Donut chart: ActBlue 63 percent versus Direct 37 percent of itemized dollars for Massachusetts House principals McGovern and Neal. Hover or focus for sources.
MA House money path — ActBlue 63%, Direct 37% (McGovern & Neal, 2026)
Campaign-night fuel — Diet Coke about three-quarters of the desk, water the rest
Reading-the-map snacks — Gummy Bears most nights, real fruit when we remember

Prepared from public election reporting. Full citations are in Sources / Notes and Predictive Model Methods.

House Forecast

October 24 AI projection, central estimate: Democrats 229, Republicans 206. The Kit GLM shared-shock national rollup is a different model and calls the chamber the other way (about 199 Democratic seats, Republican control). Prediction intervals below are the AI projection’s Democratic seats.

MetricCentral50%75%90%
House Dem seats229[225, 233][220, 237][216, 241]
House GOP seats206[202, 210][198, 215][194, 219]
Dem net House gain13[7, 19][2, 24][-2, 28]

P(Democrats win House) = 0.82. The 218-seat majority line is the reference for control. See also the 435-seat map and the House win and seat-share donuts.

House Democratic seats with 50, 75, and 90 percent prediction interval bands and a 218 majority line
435 House seats in the October 24 AI projection: 229 Democratic on the left, 206 Republican on the right. This is the chamber split, not named districts. 229 Democratic 206 Republican
October 24 AI projection. Each ring is one House seat: 229 Democratic on the left, 206 Republican on the right. This is the chamber split, not a map of named districts. The 218-seat majority line falls inside the blue block. The Kit GLM shared-shock rollup (about 199 Democratic seats, 432 seats counted) is a different model and is not drawn here.

Kit GLM versus the AI projection

Same day, two questions. The Kit GLM reallocates each state’s current House seats under a shared national shock. The AI projection is the October 24 chamber snapshot. They disagree on control.

Measure Kit GLM (shared-shock national) AI projection Gap
Dem seats (point) 199.4 (round 199) 229 −30
GOP seats (point) 232.6 (round 233) 206 +27
Chamber size in that split 432 (50-state D+R only) 435 −3
Control (point) GOP (199 < 218) Dem (229 ≥ 218) disagree
P(Dem House) ≈ 0.18 (shared shock MC) 0.82 still large
sd(Dem) ≈ 21.1 (shared); indep Poisson sd≈14.1 not used — wider under dependence
50% interval (Dem) ≈ [185, 213] [225, 233] no overlap
75% ≈ [175, 223] [220, 237] touch
90% ≈ [164, 234] [216, 241] partial overlap

The bands in this table are Democratic House seats. The Kit GLM 50% interval and the AI 50% interval do not overlap. The wider Kit GLM band uses the shared shock (sd ≈ 21.1). The independent Poisson sd ≈ 14.1 is not used. The same comparison is in the chamber notes.

Senate Forecast

Democratic caucus projected in the 49–51 range. P(Democrats win Senate) = 0.62. See also the Senate win donut.

Democratic seat intervals: 50% [49, 53]; 75% [48, 54]; 90% [46, 56].

Senate forecast: probability Democrats win House 0.82 and Senate 0.62, with Democratic caucus projected 49 to 51

Massachusetts Elections

Local and statewide races tracked alongside the national forecast. Focus races: Governor Maura Healey; MA House Lindsay Sabadosa (1st Hampshire); MA Senate Jo Comerford (Hampshire, Franklin & Worcester); U.S. House Richard Neal (MA-01). Northampton’s U.S. House district is MA-02 (Jim McGovern).

  • Maura Healey (Governor) — state campaign receipts dominate the Massachusetts filings (tens of thousands of items; dollars in the low tens of millions since 2023).
  • Lindsay Sabadosa (MA House, 1st Hampshire) — about 1,192 receipts / about $74k; high in-state share (~92.5% of receipt dollars from Massachusetts addresses).
  • Jo Comerford (MA Senate) — about 3,572 receipts / about $200k; ~96.9% Massachusetts-address dollars; Northampton-heavy itemization.
  • Richard Neal (U.S. House MA-01) — PAC-heavy federal fundraising profile; larger individual gifts ($2,000+) make up a rising share of itemized money. Included for comparison; MA-01 does not cover Northampton.
Stacked bars of ActBlue versus direct itemized individual dollars for McGovern, Neal, Warren, and Markey in 2024 and 2026
ActBlue vs direct itemized individual dollars for Massachusetts federal committees, 2024 and 2026. WinRed is $0 for these Democrats.
Line chart of observed and extrapolated ActBlue share for Jim McGovern, Richard Neal, Elizabeth Warren, and Ed Markey from 2024 through 2032
ActBlue share, observed 2024–2026 and extrapolated 2027–2032. Model extrapolation, not a causal forecast.

Criteria, family, and outcome

CriteriaFamilyOutcome
State win / Great (binary)Bernoulli logitŷ ∈ {0,1}
Great scoreGaussian identitycontinuous score
Wealthyfitted Gamma / IG LOO panelpositive $ μ + 2027–2032
HealthyBeta logitrate ∈ (0,1)
House seats (state)Poisson logper-state expected seats
House seats (national)shared logit shock (not indep Σ Poisson)out/house_seats_national_dependent.json

ActBlue vs Direct share for McGovern and Neal is in the forecast donuts. More Massachusetts charts and articles: Follow the Money · charts.

Maximum Likelihood Scenario

Late October 2026 money-flow picture drawn from public reporting:

  • Dem pad majority: late cash from Bloomberg (~$80M), House Majority PAC (~$16M), and Everytown (~$24M) concentrated on competitive House races — padding toward the low-230s rather than pure 218-seat defense.
  • GOP defensive moats: Congressional Leadership Fund and Trump-aligned groups shifting from offense (e.g. NY-03, CA-13, TX-28, IA-01, NJ-07) to defense on red-turf / open seats (e.g. MI-10, SC-01, MT-01, AZ-02, TX-23, KY-06).
  • Senate firewall: Musk-aligned super PAC spending aimed at Republican-leaning statewide races (including Texas), which can keep Senate flip odds lower even if the House moves Democratic.
  • Late redirect: House money moving from pure majority defense into expansion targets once committees treat 218 as more secure.
Maximum Likelihood Scenario chart for Democratic majority padding, Republican defensive moats, Senate firewall spending, and late House redirect to expansion seats

For Massachusetts campaign-money charts (Healey / Sabadosa / Comerford / Neal), see Massachusetts Elections and the Follow the Money charts.

Flashback: the 2020 mail-in ballot fight

In August 2020, Trump's pollsters John McLaughlin (in person) and Tony Fabrizio (by phone) met with his team at Bedminster, according to Michael Wolff's account.1 Their message was simple: because of COVID, mail-in voting would decide the election. Nearly 70 percent of voters were expected to vote early.

The pollsters admitted mail-in ballots would help Democrats. But they warned that openly fighting mail-in voting would hurt Republicans even more, because COVID would keep many people away from polling places. Their advice: tell your supporters to vote any way they can.

Trump didn't buy it. In Wolff's telling, he saw mail-in voting as part of the COVID forces lined up against him: a way for Democrats to wreck his economy and win by turning out "low-propensity" voters who usually stay home.

Why it matters for 2026: how people vote, and who makes it easy or hard, still shapes who shows up. That's part of what any forecast has to account for.

1 Michael Wolff, Landslide: The Final Days of the Trump Presidency (New York: Henry Holt and Company, 2021), 27.

Predictive Model Methods

How this October 24, 2026 forecast was assembled, checked, and published. Tools are named clearly; the steps stay in plain language.

Forecast pipeline: public sources feed NVIDIA Nemotron via Perplexity and a SQLite ETL on Jarvis (Linux); Cursor agents build the page; a human-reviewed git push deploys to Cloudflare Pages.
How the forecast is built: from public data to the live page. Git push is human in the loop.
  1. 1. NVIDIA AI forecast Seat ranges, win chances, and the money-flow narrative were produced with Perplexity’s AI forecast model perplexity/nemotron-3.5-super-120b-a12b via the ELECTIONS_FUTURE_PERPLEXITY.py run (Oct 24, 2026, 6:35 PM ET). That model read public ratings and recent news, then returned the House/Senate numbers shown above.
  2. 2. Cursor agent A Cursor cloud agent, plus a self-hosted Cursor agent on the Jarvis Linux machine, assembled the page copy, regenerated charts, checked public-facing wording, and shipped the update to Cloudflare Pages for predictiveinsightsai.com.
  3. 3. Linux OS Analysis and publish ran on Linux (Jarvis), not Windows — database loads, chart builds, git push, and Pages deploy all happened in that Linux environment.
  4. 4. SQL (Data prep) Campaign-money figures come from a the campaign-finance database pipeline (ftm.db): campaign-finance bulk / the public campaign-finance data feed loads, Massachusetts Massachusetts campaign-finance office receipts, Voteview roll-call data, and book SQL queries. ActBlue vs WinRed vs Direct on itemized individual gifts uses the conduit committee id in OTHER_ID (ActBlue C00401224; WinRed C00694323), not a memo-text grep alone.
  5. 5. Perplexity web search The AI forecast run searched the open web for late-October money and ratings context. News and data sites used for that money-flow picture include:

Hover or focus a donut chart to see the sources tied to that chart; click to pin the list and open a link. Chicago-style bibliography continues in Sources / Notes.

Sources / Notes

Chicago-style short citations for the forecast, money, and methods sections above:

  1. Decision Desk HQ. “2026 Midterm Election Forecast.” Last updated October 29, 2026, 2:25 p.m. ET. https://votes.decisiondeskhq.com/forecast/2026.
  2. Poliwave. “2026 Midterm Forecast: House and Senate Predictions.” Forecast run October 29, 2026, 8:30 p.m. ET. https://www.poliwave.com/us/fed/projection.
  3. 99chambers. “Forecast update, Oct 29, 2026.” https://99chambers.app/updates/2026-10-29.
  4. Silver, Nate, and Eli McKown-Dawson. “2026 Midterm Election Forecast: Senate & House.” Silver Bulletin, updated October 29, 2026. https://www.natesilver.net/p/nate-silver-2026-midterm-election-polls-model.
  5. Cook, Charlie. The Cook Political Report. 2026. https://www.cookpolitical.com/.
  6. Cook Political Report. “Democrats’ Projected Majority Grows as 19 More House Races Shift Away From the GOP.” October 23, 2026. https://www.cookpolitical.com/analysis/house/house-overview/democrats-projected-majority-grows-19.
  7. Cook Political Report. “2026 CPR Senate Race Ratings.” Updated October 20, 2026. https://www.cookpolitical.com/ratings/senate-race-ratings.
  8. Decision Desk HQ. “2026 Midterm Election Forecast.” October 28, 2026. https://votes.decisiondeskhq.com/forecast/2026.
  9. Cook Political Report. “A Senate Majority Is Squarely in Democrats’ Sights as Kansas Moves to Toss Up.” October 6, 2026. https://www.cookpolitical.com/analysis/senate/senate-overview/senate-majority-squarely-democrats-sights-kansas-moves-toss-two.
  10. Decision Desk HQ. “2026 Midterm Election Forecast.” October 26, 2026. https://votes.decisiondeskhq.com/forecast/2026.
  11. Federal Election Commission. “Bulk Data.” 2026. https://www.fec.gov/data/browse-data/?tab=bulk-data.
  12. Federal Election Commission. “the public campaign-finance data feed data feed.” 2026. https://api.open.fec.gov/developers/.
  13. Massachusetts Office of Campaign and Political Finance. “Public data feed.” 2026. https://api.ocpf.us.
  14. Lewis, Jeffrey B., Keith Poole, Howard Rosenthal, Adam Boche, Aaron Rudkin, and Luke Sonnet. Voteview: Congressional Roll-Call Votes Database. 2026. https://voteview.com/data.
  15. unitedstates/congress-legislators. “legislators-current.” 2026. https://unitedstates.github.io/congress-legislators/.
  16. Perplexity AI / NVIDIA AI forecast (perplexity/nemotron-3.5-super-120b-a12b). Midterm election forecast synthesis via ELECTIONS_FUTURE_PERPLEXITY.py. October 24, 2026.
  17. “Midterms Elections Live.” The New York Times, October 20, 2026. https://www.nytimes.com/live/2026/10/20/us/midterms-elections.
  18. “Mike Bloomberg Starts Spending Blitz to Help Democrats Retake House.” The Washington Post, October 20, 2026. https://www.washingtonpost.com/politics/2026/10/20/mike-bloomberg-starts-spending-blitz-help-democrats-retake-house/.
  19. “Midterm Election Fundraising / Elon Musk.” CNBC, October 16, 2026. https://www.cnbc.com/2026/10/16/midterm-election-fundraising-elon-musk.html.
  20. “Democratic Candidates and Republican Super PACs Drive Record Fundraising This Midterm.” NPR, October 19, 2026. https://www.npr.org/2026/10/19/nx-s1-6004922/democratic-candidates-and-republican-super-pacs-drive-record-fundraising-this-midterm.
  21. “Democrats’ Midterms Money Worries.” Politico, September 24, 2026. https://www.politico.com/news/2026/09/24/democrats-midterms-money-worries-disagreements-01090770.
  22. “Mike Johnson, Trump, House Races.” CNN, October 23, 2026. https://www.cnn.com/2026/10/23/politics/mike-johnson-trump-house-races.
  23. “Big Money Breaks Records, Boosts GOP: 5 Things to Know About the Midterms.” Brennan Center for Justice, 2026. https://www.brennancenter.org/our-work/analysis-opinion/big-money-breaks-records-boosts-gop-5-things-know-about-midterms.

Campaign-finance figures for Massachusetts officials are from campaign-finance bulk/data feed summaries and Massachusetts campaign-finance office receipt filings. Patterns in public records are not proof of wrongdoing.

The Declaration of Independence

“We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness.”

— Adopted July 4, 1776 by the Continental Congress

The Formal Declaration of Freedom

“That these United Colonies are, and of Right ought to be Free and Independent States; that they are Absolved from all Allegiance to the British Crown...”

— Declaration of Independence, July 4, 1776

Key Moments in the Journey to 250

July 4, 1776 The Declaration of Independence is adopted in Philadelphia. 56 delegates sign the document asserting the 13 colonies' right to be free and independent states.
1783 Treaty of Paris formally ends the Revolutionary War. Great Britain recognizes the independence of the United States.
1787–1788 U.S. Constitution is written and ratified, creating the framework of the new nation.
1791 Bill of Rights ratified — the first 10 amendments guarantee fundamental freedoms.
1803 Louisiana Purchase — the United States buys the Louisiana Territory from France, roughly doubling the size of the nation.
January 1, 1863 Emancipation Proclamation takes effect. President Abraham Lincoln declares enslaved people in the Confederate-held states free.
April 9, 1865 Surrender at Appomattox — General Robert E. Lee surrenders to General Ulysses S. Grant at Appomattox Court House, Virginia, effectively ending the Civil War.
December 6, 1865 13th Amendment ratified, abolishing slavery in the United States.
August 18, 1920 19th Amendment ratified when Tennessee approves it, barring the denial of the vote on account of sex; it is certified on August 26.
December 7, 1941 Pearl Harbor — Japan attacks the U.S. fleet in Hawaii. The United States declares war on Japan the next day and enters World War II.
June 6, 1944 D-Day — Allied forces land on the beaches of Normandy, France, in the largest seaborne invasion in history.
May 8, 1945 V-E Day — Germany's unconditional surrender ends World War II in Europe.
September 2, 1945 V-J Day — Japan formally surrenders aboard the USS Missouri in Tokyo Bay, ending World War II. President Truman had announced Japan's surrender on August 14.
May 17, 1954 Brown v. Board of Education — the Supreme Court rules unanimously that racial segregation in public schools is unconstitutional.
July 2, 1964 Civil Rights Act signed by President Lyndon B. Johnson, outlawing discrimination based on race, color, religion, sex, or national origin.
August 6, 1965 Voting Rights Act signed by President Johnson, banning racial discrimination in voting.
July 20, 1969 Moon landing — Apollo 11 astronauts Neil Armstrong and Buzz Aldrin become the first people to walk on the Moon.
September 11, 2001 September 11 attacks — terrorist attacks in New York, at the Pentagon, and in Pennsylvania kill nearly 3,000 people.
July 4, 2026 America's 250th birthday. Cities and towns across the country celebrate with parades, fireworks, and reflections on 250 years of independence.
August 17, 2026 One Monday on the National Mall — Natural History, the Washington Monument, the World War II Memorial, the Air and Space Museum, American History, the World War I Memorial, the Capitol, and Freedom Plaza.

US National Anthem - Beautiful & Emotional Instrumental Cover

A powerful and moving instrumental cover of "The Star-Spangled Banner." This arrangement is my personal tribute to the US National Anthem, blending classic patriotic spirit with a soulful touch. God Bless America.

One Monday in the Capital — August 17, 2026

Washington, D.C., is the capital of the United States. This is one Monday on the National Mall: Natural History, the Washington Monument, the World War II Memorial, the National Air and Space Museum, the National Museum of American History, the World War I Memorial, the Capitol, lunch at Aria in Federal Triangle, and Freedom Plaza.

The Monday route

Eight public stops, walked in order on August 17. Choose one to go there. The day followed museum rules and the rules of the city: posted lines, staff guidance, and the quiet a memorial asks for.

This route honors the United States — elected leaders and public servants in every office, the men and women in uniform who keep the capital and serve beyond it, and the veterans these memorials were built to remember.

01 — Morning

Natural History and the Monument

The National Museum of Natural History, then the lawn under the Washington Monument.

The Monday on the National Mall began in the National Museum of Natural History — stone walls, a cafeteria on the floor, glass elevators in the open center — and through the middle of it a megalodon, a prehistoric shark, hung on cables, jaws open, aimed at the balcony rail. For a minute the building felt like an aquarium that had forgotten the water.

I did not stay long. Out onto the Mall, the grass already warm, and onto the lawn under the Washington Monument. The obelisk rises so cleanly it almost looks unfinished — white stone against cloud, with little decoration to lean on. Flags moved in a slow circle at the base. A selfie on the lawn, and a long minute looking up. People cut across the grass toward the museums. A breeze found the flags. The rest of the Mall opened from there if I kept walking.

A life-size megalodon model hanging in the sunlit atrium of the National Museum of Natural History, jaws open above the tables on the floor
We're gonna need a bigger boat
Traveler smiling in a selfie on the lawn in front of the Washington Monument, flags around the base and clouds behind the stone
On the Monument lawn.

The megalodon, then the Monument lawn.

02 — Memorials

The World War II Memorial and the Capitol

The World War II Memorial in the morning; the Capitol’s west lawn in the afternoon.

From the Monument I walked to the World War II Memorial while the light was still high. The granite wall carries the Great Seal of the United States and the name of the war on the same stone. Around the plaza, pillars for the states stand in a ring. I stopped under Illinois, the bronze wreath in the opening, trees showing through the stone. I kept my voice low. The wall remembers men and women who wore the uniform.

This Monday on the Mall was a daytime walk, not the usual sunset route. After the museums I came east again and stood on the Capitol’s west lawn in the afternoon — metal barricades, a police line, the dome through the trees. I stayed on the public side of that line. It protects the building and the people who serve inside it. I took a full-length photo on the path, then a closer one with the dome over one shoulder. The light was flat and clear. Tour groups moved through in loose groups. I stayed long enough to see the terrace and the way the lawn opens west toward the Monument I had already stood under.

The long evening — Lincoln on the steps, Jefferson across the Tidal Basin, the Monument holding the last of the blue — is how the city likes to be remembered. Those photographs show that evening. They are not the walk I made before noon.

Granite wall at the National World War II Memorial engraved WORLD WAR II MEMORIAL above a carving of the Great Seal of the United States
World War II Memorial — the Great Seal wall.
Traveler smiling beside the Illinois granite pillar at the National World War II Memorial, a bronze wreath in the gap between the columns
Illinois pillar

Great Seal on the wall, then a stop under Illinois — mid-morning at the memorial.

Traveler standing full-body on the Capitol west lawn, metal barricades and the dome visible through the trees
West lawn — barricades, trees, the dome.
Traveler selfie on the Capitol west lawn with the dome over one shoulder
Capitol west lawn.
The Jefferson Memorial lit at dusk, reflected in the dark water of the Tidal Basin
The Jefferson Memorial at dusk. I was not there that morning.
03 — Museums

Air and Space, then American History

The National Air and Space Museum at midday, then American History a short walk west. The megalodon was already behind me.

The Smithsonian museums on the Mall are free. You walk in off the grass. I had already been under the megalodon. Early afternoon belonged to the National Air and Space Museum. Photographs stayed where the gallery welcomes them, and every case stayed as the museum presents it.

I stood in front of the Apollo lunar module — gold foil, landing legs, a flag on the crumpled skin. I posed with it because everyone does, and because the foil catches the light. The museum kept opening: the Wright brothers, Mars, more Moon hardware, and the Gemini, Cassini, and Voyager spacecraft — the long American effort to go farther, in a hall you can walk into off the grass. In the same stretch I looked up in America by Air: a 1936 Douglas DC-3 and a 1929 Ford Tri-Motor hanging from the ceiling. Then Neil Armstrong’s Apollo 11 suit — white, gold visor, glass, the plaque about one small step. School groups moved through in bright shirts. I left remembering the lunar module, the spacecraft, and the suit more clearly than any timeline on the wall.

“We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard.”

— John F. Kennedy, Rice University, September 12, 1962

A short walk west and I was in the National Museum of American History, in front of Evel Knievel’s Harley-Davidson XR-750: stars on the tank, chrome, the line about trouble beginning when you try to land it. Last museum door of the day.

The National Museum of African American History and Culture — the bronze, crown-shaped building facing the Monument — is still there to visit. So is the National Gallery of Art’s garden. I did not go inside either one on this Monday.

Traveler standing in front of the Apollo lunar module, gold foil and landing legs, at the National Air and Space Museum
Apollo lunar module
Traveler beside Neil Armstrong’s Apollo 11 spacesuit in a glass case at the National Air and Space Museum
Armstrong’s suit

The lunar module, then Armstrong’s suit — the main stops of the museum afternoon.

The bronze, crown-shaped National Museum of African American History and Culture on the National Mall
The National Museum of African American History and Culture — on the Mall, not a door I opened on August 17.
The red sandstone Smithsonian Institution Building, known as the Castle, with its towers and gardens
The Smithsonian Castle — the red sandstone building on the Mall I was already walking.
04 — Streets

The World War I Memorial and Freedom Plaza

A glass wall at Pershing Park, then Freedom Plaza while people were still on their way home from work.

After American History I was back outside. At Pershing Park, a small park near the White House, the National World War I Memorial holds a glass wall — soldiers etched in, trees and a bus reflected out. I stood back from the glass so those figures stayed the subject, then continued toward the Capitol dome.

Later I was on Pennsylvania Avenue at Freedom Plaza. First the name cut into the stone, read from the brick crosswalk; then the plaza itself, Pierre L’Enfant’s original plan of the city set into the paving, the statue of Casimir Pulaski at the edge, a crane on the skyline. I took a portrait while the light was still kind. That hour was just the avenue and the map underfoot.

The canal in Georgetown, U Street after dark, Eastern Market on Capitol Hill, the fountain at Dupont Circle — those are quieter parts of the city. They were not this Monday.

Traveler at the glass wall of the National World War I Memorial in Pershing Park, etched figures and reflections of trees
Pershing Park — the glass wall.

The glass wall at the World War I Memorial, mid-afternoon.

Freedom Plaza inscribed in stone, photographed from the brick crosswalk on Pennsylvania Avenue NW
From the crosswalk
Traveler smiling at Freedom Plaza with the inlaid street-plan paving and city buildings behind
On the paving at Freedom Plaza.

Freedom Plaza from the brick crosswalk, then on the paving itself.

Not this Monday

The C&O Canal in Georgetown at dusk, still water between stone walls and trees
The brick façade and tower of Eastern Market on Capitol Hill
The Dupont Circle fountain with its stone figures and basin

The C&O Canal in Georgetown; Eastern Market on Capitol Hill; the fountain at Dupont Circle — places I did not visit on August 17.

05 — Aria

After the Capitol: Aria

A walk west on Constitution Avenue, from the Capitol lawn to lunch at Federal Triangle.

From the Capitol I walked west. Constitution Avenue passes the Frances Perkins Building, home of the Department of Labor, and continues to the Ronald Reagan Building. Aria Pizzeria & Bar is in the food hall there, at Federal Triangle — lunch between the Capitol and Freedom Plaza.

On that walk I photographed the Grant Memorial and the Washington Monument across the lawn, then the White House from behind the barricades on Constitution Avenue. I did not photograph lunch. From there the day continued to Freedom Plaza, where the street plan is set into the paving.

The Ulysses S. Grant Memorial in the foreground, with the National Mall and the Washington Monument beyond
Grant Memorial, the Mall toward the Monument.
Looking toward the White House from behind barricades along Constitution Avenue, flags above the north lawn
Constitution Avenue, the White House behind the barricades.

West from the Capitol, along Constitution Avenue.

Blossoms

Cherry blossoms, even in August

The trees Washington is known for, even on a hot Monday.

I was not here in late March, when the trees are in bloom. You cannot walk this city, even on a hot Monday, without hearing about that season. The cherry trees around the Tidal Basin — the pool beside the Jefferson Memorial — were a 1912 gift from Tokyo. I did not walk there this Monday.

Cherry trees in bloom along the Tidal Basin, petals over the water
Cherry trees on the Tidal Basin — the picture Washington is known for, in bloom or not.
06 — Leaving

The next morning

I left the next morning. My pictures from the day end at Freedom Plaza.

I left the next morning. One Monday covers a lot, and it is still only a sketch — no last sunrise on the steps of the Lincoln Memorial, and the city keeps its own evening after you go.

Dusk on the National Mall is the picture people expect: stone picking up the last of the color, the lawns emptying. I did not stay for it.

The Washington Monument and the open lawns of the National Mall
The Lincoln Memorial at dusk, colonnade lit against a fading sky

The National Mall, which I had already crossed. The Lincoln Memorial at dusk, after I had gone.

Plan

One Monday, walked

August 17, 2026 — the day as I walked it.

Monday · August 17

From Natural History to Freedom Plaza

  1. National Museum of Natural History — megalodon hanging in the atrium.
  2. Washington Monument lawn — selfie, flags, clouds.
  3. World War II Memorial — Great Seal wall; Illinois pillar.
  4. National Air and Space — lunar module, America by Air, Armstrong’s Apollo 11 suit.
  5. National Museum of American History — Evel Knievel’s Harley-Davidson XR-750.
  6. National World War I Memorial, Pershing Park — glass wall.
  7. Capitol west lawn.
  8. Aria Pizzeria & Bar in the Reagan Building food hall at Federal Triangle.
  9. Freedom Plaza — the sign from the crosswalk, then the paving.
  10. Left the next morning.

Not visited that Monday

Not this visit

African American History and Culture

The bronze, crown-shaped museum on the National Mall. It uses timed tickets. A later morning — not this Monday.

Not this visit

The Tidal Basin at dusk

The Jefferson Memorial, the Tidal Basin, and the Lincoln Memorial after the tour buses thin out. The city’s usual evening sights.

Not this visit

Other neighborhoods

Georgetown’s canal, U Street, Eastern Market, Dupont Circle. Still there. Not this Monday.

Traveler photographs from August 17, 2026. Most Smithsonian museums on the Mall are free. Wear something you can walk in. One Monday is enough to cross the grass several times.

Symbols of American Independence

🇺🇸

The Flag

13 stripes for the original colonies. 50 stars for the states. “Old Glory” first flew in 1777.

🔔

Liberty Bell

Philadelphia. Inscribed with “Proclaim LIBERTY throughout all the Land.” Cracked but iconic.

🦅

Bald Eagle

Chosen as the national bird in 1782. Symbol of strength, freedom, and sovereignty.

🎆

Fireworks

Tradition began with the first anniversary celebrations in 1777. A burst of joy and light.

🗽

Statue of Liberty

A gift from France (1886). “Give me your tired, your poor...” — beacon of welcome and hope.

📜

The Declaration

Written primarily by Thomas Jefferson. The founding promise of equality and rights.

13+ Facts for the 250th

Founding 13 original colonies declared independence. Delaware was the first to ratify the Constitution.
Signatures 56 signers of the Declaration. The youngest was 26; the oldest was 70 (Benjamin Franklin).
Independence Day July 4th became a federal holiday in 1941, but celebrations began in 1777.
Fireworks Capital Philadelphia hosted the first official Fourth of July celebration with fireworks in 1777.
National Anthem “The Star-Spangled Banner” was written in 1814 during the War of 1812 and adopted in 1931.
250th birthday 2026 marks 250 years since the Declaration. Events are planned nationwide.

Voices of Liberty

“Give me liberty, or give me death!”
— Patrick Henry, 1775
“We must all hang together, or assuredly we shall all hang separately.”
— Benjamin Franklin, at the signing of the Declaration
“The only thing we have to fear is fear itself.”
— Franklin D. Roosevelt (echoing the spirit of resilience)
“America is a tune. It must be sung together.”
— Gerald Stanley Lee
“I am a great believer in luck, and I find the harder I work, the more I have of it.”
— Thomas Jefferson. Commonly attributed maxim reflecting his personal work ethic
An antidote to anger: “When angry count to ten before you speak. If very angry, count to one hundred.”
— Thomas Jefferson, from his “Canons of Conduct” letter to Jefferson Smith, 1825
“The state of nature has a law of nature to govern it... no one ought to harm another in his life, health, liberty, or possessions.”
— John Locke, Second Treatise of Government, 1689

Samuel Castillo Quotes Sam Adams

Worth defending: “The liberties of our Country, the freedom of our civil constitution are worth defending at all hazards: And it is our duty to defend them against all attacks.”
— Samuel Adams, “Candidus,” Boston Gazette, October 14, 1771 (The Writings of Samuel Adams, ed. Cushing, vol. 2)
Choosing freedom over comfort: “If ye love wealth better than liberty, the tranquility of servitude better than the animating contest of freedom, go home from us in peace. We ask not your counsels or arms. Crouch down and lick the hands which feed you.”
— Samuel Adams, speech at the State House of Pennsylvania, August 1, 1776
The path to independence: “The country shall be independent, and we will be satisfied with nothing short of it.”
— Samuel Adams, letter to James Warren, 1776
Awakening the public: “A true patriot would keep the attention of his fellow-citizens awake to their grievances, and not suffer them to be silenced...”
— Samuel Adams

We the People

A Tribute to Our Heroes

Here at Predictive Insights, we honor, respect, and admire our heroes in camo and navy blues — past to present. As a show of appreciation to those who devote their lives to defend this great country, we created a new art piece in their honor.

That honor extends across the public life of the United States: elected leaders and public servants in every office, the men and women in uniform, and the veterans whose service the capital’s memorials hold in trust.

Launch the Fireworks

Click or tap the canvas to celebrate 250 years of independence. (Pure client-side fireworks)

Click anywhere • Red • White • Blue

Happy Independence Day.
May the promise of 1776 continue to light the way.